OpenAI chief scientist warns AI could trick or blackmail humans, urges 'extreme caution'

OpenAI Chief Scientist Jakub Pachocki has called for "extreme caution", warning that future AI could collaborate with humans by bargaining with, tricking or blackmailing them. "No lab has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer," he said. He urged voluntary slowdowns until shared safety standards are established.